J Release Notes—V3.0-3

The main changes are the addition of recipes for Michelle nodded and chopped observations, and a rearranged directory structure for new and multi-mode instruments.

J.1 New recipes

ARRAY_TESTS

This is for Michelle. It derives a mean bias file, and calculates and reports the read noise, using the latter pair of minimum-exposure frames in the array tests sequence. Both the bias frame and readnoise are recorded in the calibration system.

NOD_CHOP

Reduction of nodded and chopped data, specifically for Michelle. Each recipe cycle comprises four frames located at two nod positions and the A and B beams. The individual beams are differenced and then so are successive pairs of observations. The differenced images are then combined into a mosaic. On successive cycles the mosaics are co-added.

NOD_CHOP_APHOT

As NOD_CHOP, but it also extracts, registers, and combines the two positive and two negative images of the source. Then it performs 3-arcsecond aperture photometry on the combined image, logging to a small text list. It compares the object name against a file of $N,Q$ standards to determine the zero point. No extinction correciton is currently applied, as the coefficients have yet to be determined.

MOVING_NOD_CHOP

As NOD_CHOP, but mosaic registration is adjusted to track the motion of an asteroid or compact comet using ephemeris data.

POL_ANGLE_NOD_CHOP

Reduces chopped and nodded poliarmetry data of point and small ($<\sim 10$ arcsec) extended sources, specifically for Michelle. The data are expects to iterate over waveplate angle before the telescope is nodded. The recipe forms integrating mosaics as NOD_CHOP, for each of four waveplate angles. For each mosaic it combines the two positive and two negative images of the source. The recipe then calculates polarisation frames and catalogues of vectors from these as earlier recipes like POL_JITTER.

POL_NOD_CHOP

As POL_ANGLE_NOD_CHOP, except the data are ordered such that the telescope performs its nodding pattern before the waveplate is turned.

J.2 Modified recipes

• ARRAY_TESTS   Instrument-specific versions for UFTI and IRCAM. Both record the read noise in the calibration system.

J.3 Global changes

The main changes from a user perspective were as follows.

• Correct data variance creation for UKIRT infra-red data. The previous calculations were for CCD data.
• Easier to switch on data-variance processing. This combined with handling of chopped data has caused some reordering of the early steps; for instance the editing of the world co-ordinate system (WCS) occurs just before dark subtraction, rather than immediately after the _IMAGING_HELLO_ preliminaries. Thus a few of the early frames such as _db and _bp now do not have a WCS defined.
• More-efficient masking using EXTRACTOR instead of PISA. Some of the acceleration is because some of the steps to sweeten the data for PISA, such as the removal of the bad pixels, are now unnecessary.
• Flat-field creation now uses an unweighted median to give approximately equal weighting to the contributing frames. The previous version of MAKEFLAT was supposed to weight the values but had bugs. The corrected task weights by the data value in the absence of a variance array assuming pure Poisson statistics, which is in appropriate for infra-red data. The unweighted median gives similar results to earlier Orac-dr versions. There is a beneficial exception where the earlier MAKEFLAT biassed towards certain contributing frames, and the unweighted median gives a more equitable division.
• An improved option for registration is available, although it is not the default. It uses the WCS to only compare sources in the overlap regions, and permits a match using a single source, provided it is within 12 pixels of the nominal WCS position.
• For moving-target registration and the ephemeris file, a bug affecting object names containing spaces is fixed.
• Better formatting of output with blank lines to block related output, and some of the commentary contain more details. The historical and unnecessary "Orac says: " prefix was removed from the commentary.
• The nearest-neighbour registration immediately prior to forming a mosaic had assumed that the brightest object with identification number 1 would be present in all the offset files. In rare cases, this may not be true. For instance, if the centred target is a faint or low surface-brightness galaxy at low galactic latitude, brighter stars relegate the galaxy to a high identification number. So the code now checks for a common-denominator object between all the frames with a higher identification number.
• The mosaic-making primitive has been partly restructured to make the code more obvious, and it also rationalises the naming. Gone is the _mu file. The mosaic with bad pixels filled has its own _fb suffix. All individual-cycle mosaics are retained. The only disadvantage is that for a single-cycle observation, there are two copies of the same mosaic, one with suffix _mos and the other with suffix _mos_0; the recipe cannot know if there is a second cycle to come.
• The mosaics have the world co-ordinate system domain set to sky so that displays with Gaia and Kappa have sky co-ordinates.
• The start and end UT times for mosaics are updated to that of the first- and last-contributing frames.
• The photometry results file now reports the sky in counts per second (previously just counts), and the exposure time. The alignment of the columns is thus slightly altered.
• The recipe tidying has been improved to remove all unwanted files. Some of the registration text files, certain suffices, and later-cycle frames were being missed.
• In the polarimetry calculations, the chip position angle is added to the offset of north with respect to the analyser, so that the vectors are also measured with respect to north. This omission had been giving vector orientations approximately 89 degrees too high for IRCAM.
• More allowance for occasional problems with the headers in the raw data, e.g. the base R.A. in degrees not hours and defunct instrument read modes.
• In earlier versions, there were fatal errors which the recipes reported, but allowed the pipeline to continue. These are now corrected so that the recipe ends its processing.
• Expanded tutorial documentation, particularly of the description of algorithms used by the primitives and more hyperlinks, and updated for the new directory structure. Various minor improvements, new hyperlinks and corrections to the recipe documentation.

The main changes from a programmer perspective were as follows.

• The primitives access user headers for steering.
• The FITS header information is accessed through generic user headers with the prefix ORAC_. This insulation permits common code for multiple instruments.
• There were upgrades for the latest versions of Kappa and ARD, notably for further use of the world co-ordinate system. In general recipes use the pixel domain for tasks like CENTROID and PSF, recording the former domain, and then resetting WCS frame after using the Kappa task.
• There is a reorganised directory structure for the recipes and primitives. These are divided into instrument-specific; general; or by topic, such as imaging or spectroscopy. The topic directories also have subdirectories for specific instruments. This restructuring permits code reuse for current and new instruments. It is also now possible to have recipes and primitives with the same name for different topics or instrument. Such scripts do have similar functions, but the exact processing depends on the data topic or some instrument attribute.